Internet of Things-Based Smart Waste Segregation System for Sustainable Waste Management in Nairobi’s Residential Estates

Abstract

Waste mismanagement remains a pressing urban challenge in Nairobi, particularly in residential estates that significantly contribute to the city’s daily waste load. Despite policy directives advocating for source-level waste segregation, manual sorting practices have proven ineffective, resulting in the landfilling of recyclable and biodegradable materials. This results in environmental pollution, health hazards, and economic losses. While global advancements in Internet of Things (IoT) and robotic technologies offer promising solutions for automating waste segregation, Nairobi is yet to integrate such innovations into its residential waste management systems. The lack of intelligent sensors, machine learning applications, and robotic mechanisms in the city’s waste practices perpetuates reliance on outdated and unsustainable methods. This study sought to bridge this technological and knowledge gap by developing and testing an IoT-based prototype designed for sustainable waste segregation in Nairobi's residential estates. The research aimed to establish the types, generation rates, and collection methods of waste in selected estates; evaluated the accuracy of sensor-driven identification in reducing contamination of recyclable materials; and designed and implemented a functional IoT-enabled segregation system. Under a sustainable framework, the study used a quantitative, applied research design that incorporates primary and secondary data sources. Field surveys were used to collect empirical data, while a prototype equipped with moisture sensors, ultrasonic, and metal detection sensors was developed and tested in a controlled environment. Statistical tools were used to assess the system's performance, including confusion matrix analysis to measure accuracy, precision, recall, and F1-score. The findings of this study demonstrated that Organic waste was the most dominant type, accounting for over 46% of household waste. Most respondents (59.3%) did not separate their waste before disposal, leading to significant contamination and inefficiencies in recycling, and rely on private waste collectors due to inconsistent public services. The prototype achieved a classification accuracy of 83%, showcasing strong potential to minimize waste contamination, enhance recycling rates, and reduce environmental degradation. The system aligns with Kenya’s broader ambitions for smart and green urban development.